I agree that it would be easier identify issues in master that we want to revert or improve.
+1 for finding out list of fixes in 1.2.x that is not in master.
For the DB fixes, I think it is OK to have tactical fixes based on adodb with vision to change to an alternative approach that gives us better story in the future. An important part of the cross-RDMS support story is to have a continuous integration that provides such checks after every checkin. If not, then we will always end up risking breaking database types that we care about.
My preference is to not change the DB layer as part of 1.3, otherwise, we will end up with something that is less likely to ship even compared to master and master-1.3.x, since this will be even more churn.